Built for
Banks & telcosRetail & e-commerceGovernmentContact centresMediaWhy Lebanese.ai
Built for dialects with little training data, starting with Lebanese, including mixed scripts like Arabizi.
The same method extends to new dialects and regions, not only Arabic.
Translation, chat and voice from the same engine, through API or ready-made widgets.
Widgets designed for 20 industries, from banking to hospitality.
Why we win
| What matters | Typical alternative | |
|---|---|---|
| Dialects | Low-resource dialects, Lebanese first, mixed with English and French | Major languages only |
| Input | Text and audio | Text only |
| IP | US patent in final approval, owned by codeX | A generic third-party API |
| Built by | Native speakers in Lebanon | Teams far from the dialect |
How it works
01
Detect
Identify the dialect and register of each message.
02
Understand
Resolve meaning, intent and local expressions.
03
Respond
Reply or translate in the right tone and language, by text or voice.
Use cases
IP & status
Lebanese.ai is protected by a US patent application for a dialect AI translation system, owned 100% by codeX, now submitted for final approval.
FAQ
Lebanese.ai is a language model for low-resource dialects, starting with Lebanese. It works in text and audio and follows Arabic, English and French mixed in one sentence.
No. Lebanese is the first dialect; the same engine is built to extend to other low-resource dialects and regions.
The Lebanese.ai dialect AI translation system, owned by codeX, has been submitted to the USPTO for final approval, and the grant is expected soon.
More from codeX
Patent pendingVerifiable AI property valuationExplore
In R&DWearable AI companionExplore
In developmentCloud platform for bespoke furniture factoriesExplore